ferrox-core 0.10.0

Core tensor ops, RoPE, GQA, and KV cache for Ferrox
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62

//! ferrox-core: tensor primitives, quantized matmul, RMSNorm, RoPE, and
//! grouped-query causal attention with a simple KV cache.
//!
//! CPU reference implementation. The op set and naming (RMSNorm, RoPE,
//! GQA, KV cache) follow the now-standard vocabulary popularized by
//! llama.cpp / vLLM / candle-transformers; the actual Rust code below is
//! written independently. See docs/THIRD_PARTY_NOTICES.md for design credit.

pub mod attention;
pub mod cache;
pub mod csa_hca_compress;
pub mod deepseek_v4_attention;
pub mod expert_store;
pub mod instance;
pub mod kernel_registry;
pub mod kv_block;
pub mod kv_disk;
pub mod kv_signature;
pub mod kv_swa;
pub mod matmul;
pub mod tensor;
pub mod threads;
pub mod turboquant;
pub mod weight_matrix;

pub use attention::{
    apply_rope_back, apply_rope_interleaved, apply_rope_interleaved_back,
    apply_rope_interleaved_with_freq_factors, apply_rope_with_freq_factors, causal_gqa_attention,
    causal_gqa_attention_paged, causal_gqa_attention_prefill,
    causal_gqa_attention_prefill_shared_kv, causal_gqa_attention_prefill_shared_kv_windowed,
    causal_gqa_attention_sinks, causal_gqa_attention_softcap, causal_gqa_attention_windowed,
    causal_gqa_attention_windowed_softcap, lightning_indexer_topk,
};
pub use cache::{
    KvBlockPool, KvCache, KvPoolExhausted, PagedKvCache, PagedKvStore, PagedStoreExhausted,
};
pub use csa_hca_compress::{channel_gated_pool, compress_block};
pub use deepseek_v4_attention::{csa_attention, hca_attention};
pub use kernel_registry::Registry as KernelRegistry;
pub use kv_block::{full_blocks, BlockHash, BlockHasher};
pub use kv_disk::{
    decode_block, encode_block, encoded_len, BlockFormatError, DiskConfig, DiskKvStore, DiskStats,
    ReadHandle, ReadOutcome, StoreError,
};
pub use kv_signature::{
    CacheSignature, KvBlock, KvDtype, SignatureError, UnverifiedBlock, BLOCK_FORMAT_VERSION,
    READABLE_FORMAT_VERSIONS,
};
pub use kv_swa::{aligned_block_size, BlockLayout, BlockLayoutError};
pub use matmul::{
    geglu, gelu, matmul_f32, rms_norm, rms_norm_per_head, silu, situ_and_mul, softcap_inplace,
    swiglu,
};
pub use tensor::Tensor;
#[cfg(feature = "cuda")]
pub use weight_matrix::cuda_dense_enabled;
#[cfg(feature = "metal")]
pub use weight_matrix::metal_dense_enabled;
pub use weight_matrix::{
    active_backend, cpu_int_dot_kind_supported, cuda_matvec_kind_supported, metal_matvec_kind_name,
    metal_mul_mm_kind_supported, BatchActs, QuantKind, WeightMatrix,
};